AI Investment Is Surging. Governments Must Protect Jobs and Deliver Better Services.
Speaking at the G20 Summit in Johannesburg on Nov 23, 2025, Singapore's Prime Minister Lawrence Wong put it plainly: the promise of AI is huge, the money is flowing into data centres and tools, and the public sector needs to put AI to work responsibly. But the real test is whether people keep their livelihoods and see better services.
Key points from PM Lawrence Wong
- AI can improve lives and productivity across the economy, but leaders should look past the hype and focus on long-term value.
- The public sector should use AI to streamline processes, strengthen service delivery, and free officers to do higher-value work.
- Businesses-from micro firms to multinationals-are already deploying AI, from off-the-shelf tools to in-house models that redesign workflows.
- Regulatory sandboxes and test beds help companies experiment safely and bring new solutions to market faster.
- The immediate responsibility for governments: address job and wage concerns, reskill and upskill workers, and support transitions into better roles.
- Example from Singapore's ports: automation lets crane operators work remotely in air-conditioned rooms, improving productivity and pay.
- The G20 can coordinate efforts so AI lifts global growth and creates good jobs, not just efficiency gains.
Why this matters to public officials
Budgets are tight and expectations are high. Citizens judge value by faster responses, fairer outcomes, and visible gains in safety, health, and livelihoods. AI is now a line item in infrastructure and workforce plans, not a side experiment.
What agencies can do now
- Define 3-5 service outcomes where AI can improve speed, accuracy, or access within 6-12 months.
- Create a lightweight AI risk and assurance process (data privacy, bias checks, human oversight, audit logs).
- Stand up a public sector AI enablement team: product owner, data engineer, model evaluator, and change manager.
- Publish procurement templates for AI tools and model services to shorten vendor onboarding.
- Run sandbox programs with clear guardrails so teams can trial models with synthetic or de-identified data.
- Build a workforce plan: which tasks will be automated, which roles expand, and which new roles you'll hire for.
- Co-design upskilling with unions and employers; fund short, stackable credentials tied to wage progression.
- Offer transition support for roles at risk: training stipends, job matching, and time-boxed redeployment pathways.
- Set productivity and service KPIs upfront; report wins and misses openly to maintain trust.
- Coordinate compute and data centre needs with sustainability targets and local job strategies.
Building worker confidence
Be straight about which tasks will change. Show the career ladder: the new skills, the timeline, and the pay outcomes. Pair every automation initiative with a funded training path and a clear human-in-the-loop policy.
Support for businesses working with government
Small firms should start with secure, AI-enabled office tools and simple workflow automations. Larger suppliers can pilot domain-specific models inside agency sandboxes, with shared evaluation metrics and red-team tests before scaling.
International coordination matters
Standards reduce risk and duplication. The OECD AI Principles and the ILO's work on decent work and the future of jobs are practical anchors for policy, assurance, and workforce transitions.
Bilateral momentum on the sidelines
Alongside the G20 sessions, PM Wong met with leaders from the EU, Finland, France, Germany, Ireland, Italy, Türkiye, and Vietnam. These ties can speed up cross-border standards, trade, and skills partnerships-especially where AI meets logistics, manufacturing, and public services.
Useful training resources
If you're planning workforce programs or pilot teams, a curated course list by role can save time. See AI courses by job function for structured paths you can adapt into civil service training.
The bottom line
AI investment without worker outcomes will backfire. Pair deployment with clear guardrails, real skills programs, and measurable service gains. Do that, and AI becomes an engine for better government and better jobs-exactly the point made in Johannesburg.
Your membership also unlocks:
